構式驅使在語義中的呈現：漢語評價性程度構式之研究

Abstract

傳統詞彙語意學研究的重心是動詞的語法和其語義的表現，對於漢語的程度副詞與形容詞或名詞互相搭配的相關研究則略顯狹隘 (參見Zhu 1956, 1982, Tang 2000ab, Liu 2002)。為了填補過往研究的不足，本文將主要探討兩個高頻率程度副詞：「很」和「有點」與其鄰近成分的分佈關係，進而提出一套以語言學角度為本的解釋。為了觀察程度副詞和其共現的相關詞類分佈，我們分析了大量的書面語料以及口語的語料，諸如中研院漢語平衡語料庫、Google搜尋引擎、聯合新聞網、以及10篇言談語料（長約139分鐘）。以往研究多認為：漢語程度副詞傾向於和具有程度性的形容詞或心理動詞共現，並且不能和無程度性的形容詞或重疊形容詞一起使用 (參見 Zhu 1956, 1982, Lü 1980, Tang 2000ab, Chui 2000a, Zhang 2002, Liu 2005討論)。然而，從實際語料中，我們可以發現和程度副詞配搭的詞類十分多樣：名詞組(嘴唇很香腸)、形容詞組 (他很藍/綠)、動詞組(很高興)、副詞組(很有點害怕/妒忌)、定冠詞組(場面很那個)、疑問詞(很怎樣)等。這些語言現象暗示我們傳統的分析似乎有其修正的空間，故本研究採用構式語法理論(Goldberg 1995, Jackendoff 1997a)重新定義漢語的程度副詞為一個形式([Degree ADV+ X] 結構)和語意(給予程度屬性性質)搭配的配對。在這個評價格式中，程度副詞為該構式的中心語(constructional head)，其功能為驅使(coerce)格式中處於X位置的詞類(grammatical category)，讓該詞類和構式能和諧地互動--亦即，X本身的語義與格式運作後所產生的相關語義是相容的。 另外，為了解決構式驅使後所造成的可能岐義性，「衍生詞法」(Pustejovsky 1995)亦在此研究中被運用。藉由此種分析的方式，我們可以得到以下兩點初步的推論（此為評價性程度格式的優點）： 1) 透過概念結構的操作，例如「語意延伸(metaphorical extension)」或「程度的層級性 (gradability)」(參見Lakoff and Johnson 1980 和Claudi and Heine 1986)以及「衍生詞法」中「經驗知識結構 (qualia structure)」，能與程度副詞共現的詞類是多樣性的，只要這些詞類本身具備語義的內涵 (semantic content)，理論上他們都可以出現在這個格式當中。 2) 「評價性程度構式」(Evaluative Degree Predication Construction)」的語言分析模式明顯比傳統的詞彙分析模式據具有優勢，因為對於這種多樣的程度副詞和其後詞類的配搭能以較經濟的(economical)方式來解釋而不需要另外加一些詞彙規則來限制，這種分析除了避免增加額外的詞義給普通的名詞組(尤其是創新性的用法(novel expressions))、也允許語意延伸的創造性、由語境來界定無層級性的形容詞、還可以解釋動詞組的各種變化、解決兩個程度副詞並列時的情況以及提供定冠詞組或疑問詞，從語境所引導出的舊訊息或可能資訊。

本研究認為應將結構意義視為詞彙的一部份來討論語意的屬性(Goldberg 1995, Jackendoff 1997a, 2002, Liu 2005)。另外，在考慮格式語法和經驗知識結構的運用前提下，語境或文化定義的訊息可以在這個結構中被詮釋。以構式語法理論為本可以用較具經濟性的方式來重新思考漢語程度副詞與詞類的配搭關係。

In recent studies of Lexical Semantics, most attention is paid to the behaviors of verbs (see Levin 1993, Tang 2000, Liu 2002 and many others). In general, few works are concerned with the correlative behaviors of degree adverbs and adjectives or nouns overall. To restore the missing pieces in lexical semantic research, this paper focuses specifically on the collocations of degree adverbs with various syntactic categories. In order to study how degree adverbs (i.e., hen ‘very’ and youdian ‘a little’) co-occur with their possible post-adverbial elements in language uses, we explore large amounts of corpora (Sinica, Google and udndata.com) and conversation data (139 minutes) and find that the assumption of previous studies on degree adverbs (Zhu 1956, 1982, Lü 1980, Tang 2000, Zhang 2002, Liu 2005) should be re-considered. Traditionally, previous woks assume that all Mandarin degree adverbs tend to occur before gradable or psych-verbs and they cannot collocate with non-gradable or reduplicative adjectives. Distinct from the traditional assumptions, an unexpectedly wide range of grammatical categories, such as NP (很香腸hen xiangchang ‘very sausage-like’), AP (很藍/綠hen lan/lu ‘KMT/DPP-like’), VP (很高興hen gaoxing ‘very happy’), ADVP (很有點害怕hen youdian haipa ‘quite sort of afraid’), DP (場面很那個changmian hen nage ‘very that (erotic)’) and interrogatives (很怎樣hen zenyang ‘very that sort of’), can occur in the X slot of [Degree ADV+ X] construction. Moreover, degree adverbs can co-occur with negative reduplicative adjectives due to pragmatic reasons (Levinson 1983, Lasersohn 1999). The observed collocations turn to the traditional question, “why do adjectives in Chinese declarative sentences have to take a degree adverb hen into a constructional premise?” and raise a new issue, “why may NP, VP, ADVP, DP or Interrogative words all collocate with the degree adverb hen?”. To deal with the question, it is proposed that the sequence of [Degree ADV+ X] is viewed as a unique construction following the approach of Construction Grammar (Goldberg 1995, Jackendoff 1997a). In other words, the form Evaluative Degree Predication Construction ([Degree ADV+ X]) is associated with the semantic interpretation ‘degree evaluation of scalar attribute or quality’ (abbreviated as EDPC). The overt marker, degree adverb, is taken as the constructional head that signals or coerces degree evaluation meaning ‘scalar-implicating quality’. Furthermore, to deal with the potential ambiguity evoking from the construction-coerced meaning, a compositional approach termed Qualia Structure in Pustejovsky (1995) is also applied. As a preliminary work on degree predications, this study has two implications: 1) A reconsideration of degree modification is needed. With the manipulation of the conceptual schema (e.g., metaphorical extension or gradability of scalar quality) and qualia roles in qualia structure, the collocations of degree adverbs with various grammatical categories in daily uses are more diverse than what is traditionally thought. The premise is that as long as the grammatical categories inherently have semantic contents, they can occur in the construction theoretically. 2) The coercive semantic interpretation can be specified in the postulation of EDPC, which has a number of merits over the lexical-rule approach: (a) It may avoid the unnecessary addition of attributive senses for nominal predicate, especially novel expressions. (b) It may allow creative sense extensions of certain contextually-defined lexically non-gradable attributions. (c) It may account for the various verbal predications. (d) It may explicate the juxtaposition of more than one degree adverb (e.g., [Degree-ADV [Degree-ADV+ AP]] (cf. Zhang 2002)). (e) It may solve the problem that demonstratives and interrogatives are also used with a degree modifier to recall a given attributive quality extracted from contexts. Ultimately, the constructional inferences discussed in this study suggest a new perspective when exploring semantic representation of Mandarin degree adverbs and the post-adverbial categories: (i) the semantic representation of scalar implicating qualities should also include constructional entities as part of lexical items (see Goldberg 1995, Jackendoff 1997a, 2002, Liu 2005); (ii) the mutual applications of constructional approach and qualia structure may help to resolve the potential ambiguity (the contextually or culturally-defined information) arising from the construction.